webscout 8.2.7__py3-none-any.whl → 8.2.8__py3-none-any.whl

webscout/Provider/turboseek.py

@@ -0,0 +1,266 @@
+from curl_cffi.requests import Session
+from curl_cffi import CurlError
+import json
+
+from webscout.AIutel import Optimizers
+from webscout.AIutel import Conversation
+from webscout.AIutel import AwesomePrompts, sanitize_stream # Import sanitize_stream
+from webscout.AIbase import Provider
+from webscout import exceptions
+from typing import Optional, Union, Any, AsyncGenerator, Dict
+from webscout.litagent import LitAgent
+
+class TurboSeek(Provider):
+    """
+    This class provides methods for interacting with the TurboSeek API.
+    """
+    AVAILABLE_MODELS = ["Llama 3.1 70B"]
+
+    def __init__(
+        self,
+        is_conversation: bool = True,
+        max_tokens: int = 600,
+        timeout: int = 30,
+        intro: str = None,
+        filepath: str = None,
+        update_file: bool = True,
+        proxies: dict = {},
+        history_offset: int = 10250,
+        act: str = None,
+        model: str = "Llama 3.1 70B" # Note: model parameter is not used by the API endpoint
+    ):
+        """Instantiates TurboSeek
+
+        Args:
+            is_conversation (bool, optional): Flag for chatting conversationally. Defaults to True.
+            max_tokens (int, optional): Maximum number of tokens to be generated upon completion. Defaults to 600.
+            timeout (int, optional): Http request timeout. Defaults to 30.
+            intro (str, optional): Conversation introductory prompt. Defaults to None.
+            filepath (str, optional): Path to file containing conversation history. Defaults to None.
+            update_file (bool, optional): Add new prompts and responses to the file. Defaults to True.
+            proxies (dict, optional): Http request proxies. Defaults to {}.
+            history_offset (int, optional): Limit conversation history to this number of last texts. Defaults to 10250.
+            act (str|int, optional): Awesome prompt key or index. (Used as intro). Defaults to None.
+        """
+        # Initialize curl_cffi Session
+        self.session = Session()
+        self.is_conversation = is_conversation
+        self.max_tokens_to_sample = max_tokens
+        self.chat_endpoint = "https://www.turboseek.io/api/getAnswer"
+        self.stream_chunk_size = 64
+        self.timeout = timeout
+        self.last_response = {}
+        self.headers = {
+            "accept": "*/*",
+            "accept-encoding": "gzip, deflate, br, zstd",
+            "accept-language": "en-US,en;q=0.9,en-IN;q=0.8",
+            "content-type": "application/json",
+            "dnt": "1",
+            "origin": "https://www.turboseek.io",
+            "priority": "u=1, i",
+            "referer": "https://www.turboseek.io/?ref=taaft&utm_source=taaft&utm_medium=referral",
+            "sec-ch-ua": '"Not)A;Brand";v="99", "Microsoft Edge";v="127", "Chromium";v="127"',
+            "sec-ch-ua-mobile": "?0",
+            "sec-ch-ua-platform": '"Windows"',
+            "sec-fetch-dest": "empty",
+            "sec-fetch-mode": "cors",
+            "sec-fetch-site": "same-origin",
+            "user-agent": LitAgent().random(),
+        }
+
+        self.__available_optimizers = (
+            method
+            for method in dir(Optimizers)
+            if callable(getattr(Optimizers, method)) and not method.startswith("__")
+        )
+        # Update curl_cffi session headers and proxies
+        self.session.headers.update(self.headers)
+        self.session.proxies = proxies # Assign proxies directly
+        Conversation.intro = (
+            AwesomePrompts().get_act(
+                act, raise_not_found=True, default=None, case_insensitive=True
+            )
+            if act
+            else intro or Conversation.intro
+        )
+        self.conversation = Conversation(
+            is_conversation, self.max_tokens_to_sample, filepath, update_file
+        )
+        self.conversation.history_offset = history_offset
+
+    @staticmethod
+    def _turboseek_extractor(chunk: Union[str, Dict[str, Any]]) -> Optional[str]:
+        """Extracts content from TurboSeek stream JSON objects."""
+        if isinstance(chunk, dict) and "text" in chunk:
+            return chunk.get("text") # json.loads already handles unicode escapes
+        return None
+
+    def ask(
+        self,
+        prompt: str,
+        stream: bool = False,
+        raw: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> dict:
+        """Chat with AI
+
+        Args:
+            prompt (str): Prompt to be send.
+            stream (bool, optional): Flag for streaming response. Defaults to False.
+            raw (bool, optional): Stream back raw response as received. Defaults to False.
+            optimizer (str, optional): Prompt optimizer name - `[code, shell_command]`. Defaults to None.
+            conversationally (bool, optional): Chat conversationally when using optimizer. Defaults to False.
+        Returns:
+           dict : {}
+        ```json
+        {
+           "text" : "How may I assist you today?"
+        }
+        ```
+        """
+        conversation_prompt = self.conversation.gen_complete_prompt(prompt)
+        if optimizer:
+            if optimizer in self.__available_optimizers:
+                conversation_prompt = getattr(Optimizers, optimizer)(
+                    conversation_prompt if conversationally else prompt
+                )
+            else:
+                raise Exception(
+                    f"Optimizer is not one of {self.__available_optimizers}"
+                )
+
+        payload = {
+            "question": conversation_prompt,
+            "sources": []
+        }
+
+        def for_stream():
+            try: # Add try block for CurlError
+                # Use curl_cffi session post with impersonate
+                response = self.session.post(
+                    self.chat_endpoint, 
+                    json=payload, 
+                    stream=True, 
+                    timeout=self.timeout,
+                    impersonate="chrome120", # Try a different impersonation profile
+                )
+                if not response.ok:
+                    raise exceptions.FailedToGenerateResponseError(
+                        f"Failed to generate response - ({response.status_code}, {response.reason}) - {response.text}"
+                    )
+
+                streaming_text = ""
+                # Use sanitize_stream with the custom extractor
+                processed_stream = sanitize_stream(
+                    data=response.iter_content(chunk_size=None), # Pass byte iterator
+                    intro_value="data:",
+                    to_json=True,     # Stream sends JSON
+                    content_extractor=self._turboseek_extractor, # Use the specific extractor
+                    yield_raw_on_error=False # Skip non-JSON lines or lines where extractor fails
+                )
+
+                for content_chunk in processed_stream:
+                    # content_chunk is the string extracted by _turboseek_extractor
+                    if content_chunk and isinstance(content_chunk, str):
+                        streaming_text += content_chunk
+                        self.last_response.update(dict(text=streaming_text)) # Update last_response incrementally
+                        yield dict(text=content_chunk) if not raw else content_chunk # Yield dict or raw string
+
+                # Update conversation history after stream finishes
+                if streaming_text: # Only update if content was received
+                    self.conversation.update_chat_history(
+                        prompt, streaming_text # Use the fully aggregated text
+                    )
+            except CurlError as e: # Catch CurlError
+                raise exceptions.FailedToGenerateResponseError(f"Request failed (CurlError): {e}")
+            except Exception as e: # Catch other potential exceptions
+                raise exceptions.FailedToGenerateResponseError(f"An unexpected error occurred ({type(e).__name__}): {e}")
+
+
+        def for_non_stream():
+            # Aggregate the stream using the updated for_stream logic
+            full_text = ""
+            try:
+                # Ensure raw=False so for_stream yields dicts
+                for chunk_data in for_stream():
+                    if isinstance(chunk_data, dict) and "text" in chunk_data:
+                        full_text += chunk_data["text"]
+                    elif isinstance(chunk_data, str): # Handle case where raw=True was passed
+                        full_text += chunk_data
+            except Exception as e:
+                raise exceptions.FailedToGenerateResponseError(f"Failed to aggregate non-stream response: {e}") from e
+            # last_response and history are updated within for_stream
+            # Ensure last_response reflects the complete aggregated text
+            self.last_response = {"text": full_text} 
+            return self.last_response
+
+        return for_stream() if stream else for_non_stream()
+
+    def chat(
+        self,
+        prompt: str,
+        stream: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> str:
+        """Generate response `str`
+        Args:
+            prompt (str): Prompt to be send.
+            stream (bool, optional): Flag for streaming response. Defaults to False.
+            optimizer (str, optional): Prompt optimizer name - `[code, shell_command]`. Defaults to None.
+            conversationally (bool, optional): Chat conversationally when using optimizer. Defaults to False.
+        Returns:
+            str: Response generated
+        """
+
+        def for_stream():
+            for response in self.ask(
+                prompt, True, optimizer=optimizer, conversationally=conversationally
+            ):
+                yield self.get_message(response)
+
+        def for_non_stream():
+            return self.get_message(
+                self.ask(
+                    prompt,
+                    False,
+                    optimizer=optimizer,
+                    conversationally=conversationally,
+                )
+            )
+
+        return for_stream() if stream else for_non_stream()
+
+    def get_message(self, response: dict) -> str:
+        """Retrieves message only from response
+
+        Args:
+            response (dict): Response generated by `self.ask`
+
+        Returns:
+            str: Message extracted
+        """
+        assert isinstance(response, dict), "Response should be of dict data-type only"
+        # Unicode escapes are handled by json.loads within sanitize_stream
+        return response.get("text", "") 
+
+if __name__ == '__main__':
+    # Ensure curl_cffi is installed
+    from rich import print
+    try: # Add try-except block for testing
+        ai = TurboSeek(timeout=60)
+        print("[bold blue]Testing Stream:[/bold blue]")
+        response_stream = ai.chat("yooooooooooo", stream=True)
+        for chunk in response_stream:
+            print(chunk, end="", flush=True)
+        # Optional: Test non-stream
+        # print("[bold blue]Testing Non-Stream:[/bold blue]")
+        # response_non_stream = ai.chat("What is the capital of France?", stream=False)
+        # print(response_non_stream)
+        # print("[bold green]Non-Stream Test Complete.[/bold green]")
+
+    except exceptions.FailedToGenerateResponseError as e:
+        print(f"\n[bold red]API Error:[/bold red] {e}")
+    except Exception as e:
+        print(f"\n[bold red]An unexpected error occurred:[/bold red] {e}")

webscout/Provider/typefully.py

@@ -0,0 +1,202 @@
+from typing import Optional, Union, Any, Dict
+import re
+from uuid import uuid4
+
+from webscout.AIutel import Optimizers
+from webscout.AIutel import Conversation
+from webscout.AIutel import AwesomePrompts, sanitize_stream
+from webscout.AIbase import Provider
+from webscout import exceptions
+from webscout.litagent import LitAgent
+from curl_cffi.requests import Session
+from curl_cffi import CurlError
+
+class TypefullyAI(Provider):
+    AVAILABLE_MODELS = ["openai:gpt-4o-mini", "openai:gpt-4o", "anthropic:claude-3-5-haiku-20241022", "groq:llama-3.3-70b-versatile"]
+
+    def __init__(
+        self,
+        is_conversation: bool = True,
+        max_tokens: int = 600,
+        timeout: int = 30,
+        intro: str = None,
+        filepath: str = None,
+        update_file: bool = True,
+        proxies: dict = {},
+        history_offset: int = 10250,
+        act: str = None,
+        system_prompt: str = "You're a helpful assistant.",
+        model: str = "openai:gpt-4o-mini",
+    ):
+        self.session = Session()
+        self.is_conversation = is_conversation
+        self.max_tokens_to_sample = max_tokens
+        self.api_endpoint = "https://typefully.com/tools/ai/api/completion"
+        self.timeout = timeout
+        self.last_response = {}
+        self.system_prompt = system_prompt
+        self.model = model
+        self.output_length = max_tokens
+        self.agent = LitAgent()
+        self.headers = {
+            "authority": "typefully.com",
+            "accept": "*/*",
+            "accept-encoding": "gzip, deflate, br, zstd",
+            "accept-language": "en-US,en;q=0.9",
+            "content-type": "application/json",
+            "dnt": "1",
+            "origin": "https://typefully.com",
+            "referer": "https://typefully.com/tools/ai/chat-gpt-alternative",
+            "sec-ch-ua": '"Microsoft Edge";v="135", "Not-A.Brand";v="8", "Chromium";v="135"',
+            "sec-ch-ua-mobile": "?0",
+            "sec-ch-ua-platform": '"Windows"',
+            "user-agent": self.agent.random()
+        }
+        self.__available_optimizers = (
+            method
+            for method in dir(Optimizers)
+            if callable(getattr(Optimizers, method)) and not method.startswith("__")
+        )
+        self.session.headers.update(self.headers)
+        self.session.proxies = proxies
+        Conversation.intro = (
+            AwesomePrompts().get_act(
+                act, raise_not_found=True, default=None, case_insensitive=True
+            )
+            if act
+            else intro or Conversation.intro
+        )
+        self.conversation = Conversation(
+            is_conversation, self.max_tokens_to_sample, filepath, update_file
+        )
+        self.conversation.history_offset = history_offset
+
+    @staticmethod
+    def _typefully_extractor(chunk: Union[str, Dict[str, Any]]) -> Optional[str]:
+        if isinstance(chunk, str):
+            if isinstance(chunk, bytes):
+                chunk = chunk.decode('utf-8', errors='replace')
+            match = re.search(r'0:"(.*?)"', chunk)
+            if match:
+                content = match.group(1).encode().decode('unicode_escape')
+                return content.replace('\\', '\\').replace('\\"', '"')
+        return None
+
+    def ask(
+        self,
+        prompt: str,
+        stream: bool = False,
+        raw: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> Dict[str, Any]:
+        conversation_prompt = self.conversation.gen_complete_prompt(prompt)
+        if optimizer:
+            if optimizer in self.__available_optimizers:
+                conversation_prompt = getattr(Optimizers, optimizer)(
+                    conversation_prompt if conversationally else prompt
+                )
+            else:
+                raise Exception(
+                    f"Optimizer is not one of {self.__available_optimizers}"
+                )
+        payload = {
+            "prompt": conversation_prompt,
+            "systemPrompt": self.system_prompt,
+            "modelIdentifier": self.model,
+            "outputLength": self.output_length
+        }
+        def for_stream():
+            try:
+                response = self.session.post(
+                    self.api_endpoint, 
+                    headers=self.headers, 
+                    json=payload, 
+                    stream=True, 
+                    timeout=self.timeout,
+                    impersonate="chrome120"
+                )
+                if not response.ok:
+                    raise exceptions.FailedToGenerateResponseError(
+                        f"Failed to generate response - ({response.status_code}, {response.reason}) - {response.text}"
+                    )
+                streaming_text = ""
+                processed_stream = sanitize_stream(
+                    data=response.iter_content(chunk_size=None),
+                    intro_value=None,
+                    to_json=False,
+                    content_extractor=self._typefully_extractor,
+                )
+                for content_chunk in processed_stream:
+                    if content_chunk and isinstance(content_chunk, str):
+                        streaming_text += content_chunk
+                        yield content_chunk if raw else dict(text=content_chunk)
+                self.last_response.update(dict(text=streaming_text))
+                self.conversation.update_chat_history(
+                    prompt, self.get_message(self.last_response)
+                )
+            except CurlError as e:
+                raise exceptions.FailedToGenerateResponseError(f"Request failed (CurlError): {e}")
+            except Exception as e:
+                raise exceptions.FailedToGenerateResponseError(f"An unexpected error occurred ({type(e).__name__}): {e}")
+        def for_non_stream():
+            for _ in for_stream():
+                pass
+            return self.last_response 
+        return for_stream() if stream else for_non_stream()
+
+    def chat(
+        self,
+        prompt: str,
+        stream: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> str:
+        def for_stream():
+            for response in self.ask(
+                prompt, True, optimizer=optimizer, conversationally=conversationally
+            ):
+                yield self.get_message(response)
+        def for_non_stream():
+            return self.get_message(
+                self.ask(
+                    prompt,
+                    False,
+                    optimizer=optimizer,
+                    conversationally=conversationally,
+                )
+            )
+        return for_stream() if stream else for_non_stream()
+
+    def get_message(self, response: dict) -> str:
+        assert isinstance(response, dict), "Response should be of dict data-type only"
+        text = response.get("text", "")
+        try:
+            formatted_text = text.replace('\\n', '\n').replace('\\n\\n', '\n\n')
+            return formatted_text
+        except Exception:
+             return text
+
+if __name__ == "__main__":
+    print("-" * 80)
+    print(f"{'Model':<50} {'Status':<10} {'Response'}")
+    print("-" * 80)
+    working = 0
+    total = len(TypefullyAI.AVAILABLE_MODELS)
+    for model in TypefullyAI.AVAILABLE_MODELS:
+        try:
+            test_ai = TypefullyAI(model=model, timeout=60)
+            response_stream = test_ai.chat("Say 'Hello' in one word", stream=True)
+            response_text = ""
+            for chunk in response_stream:
+                response_text += chunk
+            if response_text and len(response_text.strip()) > 0:
+                status = "OK"
+                clean_text = response_text.strip()
+                display_text = clean_text[:50] + "..." if len(clean_text) > 50 else clean_text
+            else:
+                status = "FAIL (Stream)"
+                display_text = "Empty or invalid stream response"
+            print(f"\r{model:<50} {status:<10} {display_text}")
+        except Exception as e:
+            print(f"\r{model:<50} {'FAIL':<10} {str(e)}")